Abstract
What is belief and how does it relate to certainty? This paper takes on some of the strongest arguments for views that identify belief with certainty, credence 1 or maximal confidence. It considers an influential version of the position on which the assignment of credences is context-dependent in crucial ways (especially Clarke [2013]), arguing that such a position is not viable. Examining these arguments and approaches in detail is shown to illuminate some key issues about credences and beliefs and the relation between them, as well as elucidating the role of representation in the understanding of beliefs. While rejecting the position that belief is certainty may seem to reopen the threat of arbitrariness and imprecision, I argue that it is essential to our understanding of beliefs to recognise that our belief-talk is vague. In the final section, I explore the nature and role of doxastic states that are borderline beliefs, which I call vague beliefs. Acknowledging the vagueness of the category of beliefs is necessary to reflect the rich and complex pattern of our doxastic states and commitments in the inevitable absence of certainty.